A strange thought occurred to me recently.

Maybe artificial intelligence is finally solving a problem humanity has been trying to solve for more than a century — just not in the way we expected.

People have long imagined a world where national and linguistic borders matter less.

Cosmopolitanism offered the big philosophical version: a person could see themselves not only as a citizen of a country or member of a community, but as part of humanity as a whole.

Esperanto offered something closer to an engineering solution. If languages divide us, create one simple, neutral language that everyone can learn.

It was an elegant idea.

It also had one practical problem: for a common language to work, huge numbers of people first had to agree to learn it.

Humanity has never been especially good at projects that begin with, “First, everyone has to agree.”

Then came the internet.

It really did connect the world. Someone in Montreal can write to someone in Tokyo in seconds. A scientist in Tbilisi can work with a colleague in Boston.

But being able to reach someone is not the same as being able to speak to them naturally.

Language remained a barrier.

And not always a fair one.

You can be excellent at what you do, have strong ideas and years of experience, and still sound less convincing simply because you are writing in a language that is not your own.

The idea may be good.

You used the wrong article.

Suddenly, you sound a little less impressive.

This is where I think AI changes the rules in a more fundamental way.

Not because it gives us another common language.

Because a common language may no longer be necessary.

You write in Russian.

Someone else writes in French.

A third person writes in Japanese.

And it matters less and less whether the person on the other side actually speaks your language.

This is no longer just translation. We have had translation tools for years.

The interesting part is that you can now ask:

“Translate this into English.”

But you can also ask:

“Express this the way someone in Canada would naturally say it.”

That is a different task.

You can think in the language in which you think best, while communicating with almost anyone.

In that sense, AI offers almost the opposite of Esperanto.

Esperanto said:

Let’s all learn one language.

AI says:

You don’t have to.

Keep yours. I’ll try to help you understand each other.

That strikes me as a much more interesting shift than simply having better translation software.

Most earlier attempts at universality required people to adapt themselves to the system: learn the shared language, learn the rules, learn somebody else’s way of communicating.

Here, the system is trying to adapt itself to the person.

Of course, there is another side to this.

If two people do not know each other’s languages and communicate through AI, they are also trusting it to interpret them.

That is a considerable amount of power.

Someone writes something harsh, and the system may soften it.

Something is ambiguous, and the system chooses one meaning.

A joke only makes sense inside one culture, and the system decides how to explain it to another.

So there is a useful question hiding inside all of this.

Are we actually beginning to understand one another better?

Or are we simply getting a very good intermediary that makes us easier for one another to understand?

In a few years, we may stop caring what language the person on the other side of the screen speaks.

And that would be a curious historical outcome.

Esperanto tried to give humanity one common language.

Artificial intelligence is offering something else:

keep your languages.

Maybe we no longer need a common one.